Stabilizing through Poor Information

This paper studies the effect of asymmetric information on equilibrium stability in a class of linear models where the actual state depends on the forecasts about it. Stability is deffined by the so-called eductive criterion which relies on common knowledge of rationality. The main result is that stability obtains when the proportion of uninformed agents is high enough. The expectational behavior of these agents indeed displays more inertia. This behavior, and then the actual outcome, are therefore easier to predict. This result is linked to the issue of informational efficiency. Extensions to cases with higher order uncertainty, additional agents heterogeneity, and sunspots are also considered.